# (Re)Process the records
# Convert the individual marcxml_in files to raw marc and write them all to a single .mrc file
# OUTPUT FILE
marcRecsOut_orig_recs = pymarc.MARCWriter(file(aco_globals.batch_folder+'/'+batch_name+'_0_orig_recs.mrc', 'w'))
marcxml_dir = aco_globals.batch_folder+'/marcxml_in'
for filename in os.listdir(marcxml_dir):
file_path = os.path.join(marcxml_dir,filename)
if os.path.isfile(file_path):
if file_path[-3:]=='xml':
marc_xml_array = pymarc.parse_xml_to_array(file_path)
for rec in marc_xml_array:
rec = aco_functions.pad_008(rec)
rec_001 = rec.get_fields('001')[0]
print rec_001
marcRecsOut_orig_recs.write(rec)
marcRecsOut_orig_recs.close()
# Extract the OCLC numbers from each record and write records to .mrc and .txt files depending if record contains OCLC number or not
# INPUT FILE
marcRecsIn_orig_recs = pymarc.MARCReader(file(aco_globals.batch_folder+'/'+batch_name+'_0_orig_recs.mrc'), to_unicode=True, force_utf8=True)
# OUTPUT FILES
try:
os.makedirs(aco_globals.batch_folder+'/'+batch_name+'_1/')
except OSError as exception:
if exception.errno != errno.EEXIST:
raise
oclc_nums_for_export = codecs.open(aco_globals.batch_folder+'/'+batch_name+'_1/'+batch_name+'_1_oclc_nums_for_export.txt', 'w', encoding='utf-8')
all_recs_analysis_txt = codecs.open(aco_globals.batch_folder+'/'+batch_name+'_1/'+batch_name+'_1_all_recs_analysis.txt', 'w', encoding='utf8')
marcRecsOut_orig_recs = pymarc.MARCWriter(file(aco_globals.batch_folder+'/'+batch_name+'_0_orig_recs.mrc', 'w'))
# Convert the individual marcxml_in files to raw marc and write them all to a single .mrc file
marcxml_dir = aco_globals.batch_folder+'/marcxml_in'
for filename in os.listdir(marcxml_dir):
file_path = os.path.join(marcxml_dir,filename)
if os.path.isfile(file_path):
if file_path[-3:]=='xml':
marc_xml_array = pymarc.parse_xml_to_array(file_path)
for orig_rec in marc_xml_array:
orig_rec = aco_functions.pad_008(orig_rec)
orig_003_value = orig_rec.get_fields('003')[0].value() # the institutional code from the 003
orig_001_value = orig_rec.get_fields('001')[0].value() # the local BSN from the 001
orig_245 = orig_rec.get_fields('245')[0]
orig_245a = orig_245.get_subfields('a')[0] # the main title from the 245 subfield a
print orig_001_value
orig_rec_count_tot +=1
# Extract the OCLC numbers from each record and write records to .mrc and .txt files depending if record contains OCLC number or not
rec_oclc_nums = set() # set variable to capture unique list of OCLC numbers for just this record
oclc_num_exists = False
for oclc_num_field in rec.get_fields('035','079'): # iterate through all the 035/079 fields in the original partner record
oclc_num_field_az = oclc_num_field.get_subfields('a','z') # capture the list of all subfields a or z in the 035/079 fields
if len(oclc_num_field_az) > 0: # check if subfield a or z exists in the 035/079 fields
for this_az in oclc_num_field_az: # iterate through each of the subfields a or z
if this_az.startswith('(OCoLC)') or this_az.startswith('o'): # check if the subfield data is an OCLC number
'w',
encoding='utf8')
marcRecsOut_orig_recs = pymarc.MARCWriter(
file(aco_globals.batch_folder + '/' + batch_name + '_0_orig_recs.mrc',
'w'))
# Convert the individual marcxml_in files to raw marc and write them all to a single .mrc file
marcxml_dir = aco_globals.batch_folder + '/marcxml_in'
for filename in os.listdir(marcxml_dir):
file_path = os.path.join(marcxml_dir, filename)
if os.path.isfile(file_path):
if file_path[-3:] == 'xml':
marc_xml_array = pymarc.parse_xml_to_array(file_path)
for orig_rec in marc_xml_array:
orig_rec = aco_functions.pad_008(orig_rec)
orig_003_value = orig_rec.get_fields(
'003')[0].value() # the institutional code from the 003
orig_001_value = orig_rec.get_fields(
'001')[0].value() # the local BSN from the 001
orig_245 = orig_rec.get_fields('245')[0]
orig_245a = orig_245.get_subfields('a')[
0] # the main title from the 245 subfield a
print orig_001_value
orig_rec_count_tot += 1
# Extract the OCLC numbers from each record and write records to .mrc and .txt files depending if record contains OCLC number or not
rec_oclc_nums = set(
) # set variable to capture unique list of OCLC numbers for just this record
oclc_num_exists = False
for oclc_num_field in rec.get_fields(
# (Re)Process the records
# Convert the individual marcxml_in files to raw marc and write them all to a single .mrc file
# OUTPUT FILE
marcRecsOut_orig_recs = pymarc.MARCWriter(
file(aco_globals.batch_folder + '/' + batch_name + '_0_orig_recs.mrc',
'w'))
marcxml_dir = aco_globals.batch_folder + '/marcxml_in'
for filename in os.listdir(marcxml_dir):
file_path = os.path.join(marcxml_dir, filename)
if os.path.isfile(file_path):
if file_path[-3:] == 'xml':
marc_xml_array = pymarc.parse_xml_to_array(file_path)
for rec in marc_xml_array:
rec = aco_functions.pad_008(rec)
rec_001 = rec.get_fields('001')[0]
print rec_001
marcRecsOut_orig_recs.write(rec)
marcRecsOut_orig_recs.close()
# Extract the OCLC numbers from each record and write records to .mrc and .txt files depending if record contains OCLC number or not
# INPUT FILE
marcRecsIn_orig_recs = pymarc.MARCReader(file(aco_globals.batch_folder + '/' +
batch_name + '_0_orig_recs.mrc'),
to_unicode=True,
force_utf8=True)
# OUTPUT FILES
try:
os.makedirs(aco_globals.batch_folder + '/' + batch_name + '_1/')
